Once the data has been acquired by an interferometer such as the IRAM Plateau de Bure Interferometer (PdBI) data, two different approaches may be used for its reduction and analysis:
The choice of clearly separating calibration and imaging+deconvolution was taken at start of the PdBI, when the limiting number of antenna forbid the use of self-calibration. This choice turned out to be useful for several reasons: 1) It simplified the structure of each program by keeping at the minimum the complexity of the data format required at each step. 2) It ensured clear responsibility between developers. 3) It enabled well-defined break points in data reduction and analysis in steps which are more and more generic: While many points of the calibration algorithms inside CLIC are specific to PdBI data (in particular its range of Signal-to-Noise ratio), the algorithms of imaging+deconvolution can be used in many different contexts and the visualization and analysis of spectra cubes is mainly independent of the instrument that delivered the data. This last point implies that users can import data (mainly through FITS format) in MAPPING for imaging and deconvolution and in GREG for visualization and analysis. But the reverse is also true: While calibration of PdBI data should be done inside CLIC, imaging+deconvolution and visualization+analysis can be done in other softwares (e.g. MIRIAD, AIPS, AIPS++ newly called CASA for the imaging and deconvolution and KARMA for the visualization and analysis).
With the improvements of PdBI (increase of the number of antennas and better receiver sensitivities) and with the advent of a new generation of interferometer (ALMA), an additional step of self-calibration may improve the consistency of the final results by imposing additional consistent constraints on the calibration. We are now exploring this possibility in MAPPING.